I've rebuilt several of these steamboat pumps over the years here
in the UK, and they were pretty much identical.  A few observations:

The main camshaft bearings were single-split white metal sleeves with
a grease hole for the lubricator and a hole which located on a set
screw in the housing to prevent them creeping axially.  It's obviously
a mistake to turn this set screw so far in that it distorts the white
metal or touches the camshaft.  All the camshaft bearings I encountered
were dry -- because no-one had turned the lubricators for years! -- and
knocked slightly, but were fairly evenly worn and therefore serviceable.

Having cleaned everything I followed the old technique of nipping
them up with the lightest coating of assembly oil and then backing
the two housing screws off fractionally until the camshaft turned
freely.  Using new LMP grease in the lubricators I then worked it
through the bearings in the way Jeffrey Wood described, resulting in
silent, free-running bearings with no detectable play or tight spots.
Perfectionists will want to check bearing tightness again after a few
hours running.

It goes without saying that, especially if they have been disturbed,
accurate axial alignment of the bearing housings on each end of the
pump manifold must be established using card shims before doing any
of this.

The cam follower bearings were brass tubes pressed into the wooden
rollers and held in by recessed brass washers retained by countersunk
wood screws.  The silver-steel axles were somewhat worn on their cam-
ward sides which allowed the followers to "skitter", but turning these
180 degrees and using molybdenum disulphide grease rendered them
free-running yet noiseless.

While the piano isn't being played the leather facings on the cam
followers tend to develop flat spots which cause a rumbling sound.
Finding leather which is hard enough yet soft enough for quietness is
an intractable problem.  I found the silent, long-lasting solution was
to press tight-fitting polyurethane rubber tube of the correct wall
thickness onto the bare followers.  Allowing it to overhang both roller
edges slightly (the stirrup allowed just enough room) proved sufficient
location in most cases, but the tube can also be fixed with strategic
flathead pins punched below the surface as originally.

When disassembling these pumps the observant will notice that the
stirrups which hold the cam followers are shimmed with card onto their
pneumatics, not only to align each axle with its cam so the follower
runs true, but to set the correct opening span of each pneumatic to
achieve maximum, even, pulse-free suction.  It's vital that these are
re-established correctly during reassembly, no matter how exactly each
pneumatic appears to have been remade and reattached.

The card stiffeners inside the pneumatics are essential again not only
to prevent noise from any billowing of the cloth but also to maintain
pulse-free suction.  The pneumatics must be remade with exactly the
correct original span: they will not develop a high enough pump tension
if the span is increased even by a small amount.

The cam followers will of course knock if they lose contact with the
cam because the pneumatic doesn't empty fast enough under spring
tension.  The problem here is usually that the anti-slap crossbars are
too close to allow the exhaust valves to open sufficiently.  A generous
sixteenth of an inch play can be allowed without them becoming noisy.
Fortunately the exhaust valves can be adjusted once the pump is
reassembled.

Unfortunately the inlet valves can't.  So strop all the valve leathers
over something like a wooden chair back until they are flexible and
supple and, unless modern closed-cell neoprene faces are to be fitted,
rub real talc (not French chalk) into the nap.  It isn't necessary to
remove the felt backing if it's in good condition -- stropping will
fracture and relax the glue but the felt never seems to fall off.

Use one thin coat of thixotropic adhesive on the back of the leather
if replacing the felt, and to attach thin closed-cell neoprene faces
if desired.  Thixotropic adhesive won't harden or shrink, will allow the
valves to seat properly and can be damped with solvent and rolled off
at any time in future if necessary.  Before re-attaching the pneumatics
ensure that the crossbars allow ample valve lift without fouling the
inside of the moving board

Reattaching the pneumatics to the manifold of these pumps has always
been a problem without the original hot cupboards, cauls and jigs, and
I think everyone I know who has tried to rebuild one of these pumps has
failed to get a perfect alignment and seal of all six first time.

Of course I shall be vilified for saying this, but the surest basis
for getting these pumps to work perfectly is to glaze the manifold and
pneumatic faces with PVA (_not_ PVC-E!) solution and when it is dry
fix the pneumatics on with silicone RTV adhesive, which will allow
a perfect seal and alignment without jigs or clamps first time every
time.  Pumps rebuilt this way are actually quieter too.

Contrary to what others insist, silicone adhesive on a PVA glaze will
_not_ affect the original wood adversely in any way whatsoever;
properly made, the bond is guaranteed to hold tight for at least 50
years and rising (as in the oldest B-52s); the pneumatics can easily
be removed at any time without any further risk of damage merely by
pushing a thin flexible blade between the bonded faces; the inlet
valves can then be serviced as necessary;  the old RTV can be easily
rolled off the PVA surface; and the pneumatics reattached using more
RTV.  If for some unfathomable reason anyone should want to go back
to using hot glue, wiping away the PVA glaze will reveal perfect clean
wood ready for it.

Above all rebuilding a steamboat pump in this way will allow a piano
the chance to reproduce correctly -- which is the object of the
exercise.

Patrick Handscombe
Wivenhoe, Essex, UK